One of the central innovations of the health information technology (HIT) revolution is the use of electronic reminders to provide clinicians with potentially relevant information to support clinical decisions. On the one hand, electronic reminders can present helpful information to support better patient care. On the other hand, given the limits of human cognitive processing and attention, there is the distinct possibility tha reminders may overburden clinicians with too much information. There is currently little empirical evidence weighing the benefits and burdens of increasing the informational content of systems of electronic reminders. Without a better understanding of how varying the information presented in systems of HIT affects patient care, it will not be possible to build HIT systems - an enterprise with $35 billion in annual investments - to optimize patient care. The long-term goal is to develop empirical methods using data from real health care systems to identify the effect of important design margins of HIT systems on actual patient care and health. The objective in this particular application is to quantitatively measure key characteristics of variation in electronic reminder systems in 130 health care systems within the Veterans Health Administration (VHA) and then to use these measures to estimate the effect of increasing the informational content of a system of reminders on the quality of primary care delivered. Preliminary data using counts of informational items associated with reminders suggest wide variation across systems of reminders. The same provider caring for the same patient may be required to process 5 or 40 reminder-related pieces of information on preventive care and disease management, depending on the health care system and time. I will achieve this objective by pursuing three specific aims: (1) Quantify differences in sets of electronic reminders across VHA health care systems over time. In addition to preliminary evidence using counts of informational items, I will quantitativel measure other dimensions of variation (e.g., topic breadth, complexity, and comprehensibility) in the informational content of systems of reminders. (2) Estimate the dose response of primary care quality measures to the informational content of reminders, using natural experiments on the set of reminders presented to clinicians. (3) Determine how the dose response to the informational content of reminders depends on patient, provider, and organizational context. Using administrative patient- and provider-level data, external data on provider board scores, and surveys at the organizational level, I will examine how the response to reminders differs across contexts described by these characteristics. The approach is innovative in its use of detailed administrative data, variation in the content of information presented in HIT systems, and natural experiments changing this information content. The proposed research is significant because it represents a first step towards defining the right amount of information in HIT to optimize patient care, given provider cognitive limitations. Progress in this area is needed to enable the evaluation and design of HIT systems to optimize patient care.